Skanda Amarnath is the executive director of Employ America and a former hedge fund economist. He rejoins Macro Musings to talk about the fate of the Phillips Curve, the inflation outlook, the Fed’s new framework, and his vision for a better monetary policy future. David and Skanda also discuss the Fed’s flawed assessment of maximum employment, how to modify the central bank’s Summary of Economic Projections, and the significance of capacity constraints vs labor utilization.
Read the full episode transcript:
Note: While transcripts are lightly edited, they are not rigorously proofed for accuracy. If you notice an error, please reach out to [email protected].
David Beckworth: Skanda, welcome back.
Skanda Amarnath: Thanks for having me, David. Looking forward to our discussion today.
Beckworth: Absolutely. It's been fun conversing with you on Twitter, in-person, and you've made several visits to the show. I think the first time was back in Arlington, but you've done some since then. One thing that's changed since those first few visits is that your title is now executive director of Employ America, so before you were a research director. Congratulations, first off, but tell us about the change. What's going on at Employ America?
Amarnath: Thank you, thank you. So I guess I'm now the executive director. While it's certainly a promotion and a privilege, also not as much as substantively, we were all part of the same group that's helping to lead things together between, my colleagues, Sam Bell, Kim Steins, Arnab Datta and Alex Williams, so we are all in this together. Yes, the titles have shifted a bit, but it's also a little bit more… the substance has not changed as much, and we are still moving the same direction as we always have.
Beckworth: So what does Employ America do? I know I've asked this question before, but for new listeners, tell us about Employ America.
Amarnath: Sure, so we try to advocate for macroeconomic polices that can help to ensure tighter labor markets and better labor markets over time, so can we achieve higher rates of employment and better rates of wage growth over time, in a sustained manner, such that the tight labor market, let's say, of 2019, isn't just this fleeting macroeconomic moment but actually can be something that is replicated more sustainably over time with better management of monetary and fiscal policies and everything else that is encompassed by macroeconomic policy settings.
Beckworth: Yeah, I like to think of your group as trying to put into play what Karl Smith says, “give capitalism a chance.” Really run the economy hot but at a sustainable level, and really get people into the labor force, get them working, let's get full employment, so let's really try capitalism.
Amarnath: Karl is in some ways a good analog for some of our ideas in terms of, we are trying to let market economies reveal their strengths in terms of being participatory, in a way that businesses are competing for labor. That's not a bad thing. That's not anti-market. That's not anti-sort of a robust private sector, but there are certain demand policy settings that really shape whether that's happening or not. If we have really slack labor markets where that's not happening, that's also the kind of environment where you don't see a lot of investment, or you don't see the same kind of dynamism that, I think, has the chance to be replicated if we do get a more cooperative and coordinated set of policies across the board. And so we're trying to do our part. Obviously, macro policy isn't everything, but for us, maybe a little bit more so.
We are trying to let market economies reveal their strengths in terms of being participatory, in a way that businesses are competing for labor. That's not a bad thing. That's not anti-market. That's not anti-sort of a robust private sector, but there are certain demand policy settings that really shape whether that's happening or not.
Beckworth: Absolutely, good stuff you guys are putting out. In fact, you just put out a new piece with your colleague, Alex Williams, titled *Beyond the Phillips Curve: A Dynamic Approach to Communicating Assessments of 'Maximum Employment'.* I'm going to get to that, and that'll be the lion's share of our discussion today, but before we get to that, I thought we could motivate by first talking about the Fed's framework, because this is, in some ways, a critique of how the Fed's implementing its new framework. I know most of our listeners know what this framework is, but there's some new listener, again, who's just joined the show, or for the average person on the street, if they ask you, "Skanda, what is the Fed's new framework?" How do you explain it to them?
Explaining the Fed’s New Framework
Amarnath: It's either two major revisions to the Fed's framework that define it nowadays. The first that's pretty well-telegraphed was the notion that the Fed had to make up for past misses on its inflation targets. So the Fed has a 2% inflation target that it wants to achieve over time. The 2010s, that decade, was largely marked by inflation missing its target to the downside, so we've got lower inflation than 2%, and the Fed thinks it's really important to hit 2% over time, so now there is this kind of a little bit vague, flexible approach to allowing inflation to compensate for those misses.
Amarnath: That's the first part, I guess you could say, the headline thing that most market participants have fixated over, flexible average inflation targeting, so that flexible means we're not doing this in a mechanical manner, but we are trying to allow inflation to actually get above 2% to make up for all of our misses in the past. That's the one part. The second part that probably received less attention, but also was a bigger surprise in August of 2020, was the notion that the Fed was no longer going to try and land the labor market on a pin. They were not going to try and hit employment to a particular point, and then sort of optimize policy around it.
Amarnath: Before it used to be that the Fed would try to minimize deviations away from where they thought maximum employment was, whereas now they're just addressing shortfalls, so it's an asymmetric approach to aiming for particular labor market outcomes. Yes, there is a problem when the unemployment rate is really high. It's a problem when employment is really far from where it was before the pandemic. Is it a problem if we have employment that's exceeding their estimates of maximum employment? Is it really a problem if the unemployment rate is 3.5% but FOMC members think maximum employment is 4%?
Amarnath: That assessment is more contextual. It's something that we really try to highlight in our piece. The contextual nature of that is 3.5% might be inflationary, it might not. That's not actually something that's a given. We don't really know whether 3.5% unemployment was inflationary. It certainly didn't appear to be inflationary before the pandemic. Whether it's actually inflationary the next time we get to 3.5% unemployment is really not something that can be governed by a particularly systematic relationship, and the Fed has to be flexible to that. The framework seems to specify that kind of flexibility, to the Fed's credit, but the Fed's communications are still stuck in the old ways. That's part of what our piece tries to get at.
We don't really know whether 3.5% unemployment was inflationary. It certainly didn't appear to be inflationary before the pandemic. Whether it's actually inflationary the next time we get to 3.5% unemployment is really not something that can be governed by a particularly systematic relationship, and the Fed has to be flexible to that. The framework seems to specify that kind of flexibility, to the Fed's credit, but the Fed's communications are still stuck in the old ways.
Beckworth: It's interesting, you mentioned the Fed's communication is stuck in its old ways. I think many people are stuck in their old ways when thinking about inflation, this new framework, and so I think it's important to have these conversations to remind us, this is a new framework. And maybe the Fed can do a better job, as you suggest in your paper, but I think people in the markets have a hard time internalizing this as well.
Amarnath: Yeah, part of it is they've introduced this framework at a time when we've had historically volatile macroeconomic data. Where even with the disappointing jobs prints of the past couple of months, it's still historically strong. The inflation prints, according to the Consumer Price Index, have also been really strong. Oil prices were negative a little bit more than a year ago, versus now they're meaningfully significantly higher, I should say.
Amarnath: We’ve seen this big base effects, these big movements in prices, over the last 12 to 15 months. And so when you're introducing a framework about average inflation targeting, right in the middle of that, it's in the middle of the eye of the storm, it's really hard for people to make full sense of what actually am I supposed to be anchored towards. The Fed says they want to make up for past sins to the downside. But how they do so is really hard to actually calibrate and calculate if you're a market participant.
Beckworth: Absolutely, and I think it's important for politicians to understand it, as well as you and me, and other regular people in America, because I think it's easy to get freaked out. You look at the big deficits. You hear the Fed's not taking things seriously. I won't mention any names, but there's some very serious people, commentary out there, that I think completely misses what's happening about this change we are experiencing right now.
Beckworth: Let me ask a couple more questions about this framework before we move to your paper. So some of the parameters are not very clear, like how quick the fed will make up or how far back it will go, but one thing I think they did say, and I wanted to get your take on this, because I'm not sure I fully understand it, but they want to see 2% inflation for at least a year, as well as full employment, right? You can't just have... You're going above 2% for a year, and then you change policy. You could be going above 2% for a year and still not be at full employment, and you've still got to keep going, right, with easy policy? Is that right?
Amarnath: In September of 2020, that FOMC meeting, the first FOMC meeting after they released their new framework. They released forward guidance about what it would take for the Fed to start considering interest rate hikes. In that statement, there were three components to it. One was that the labor market had reached conditions consistent with FOMC members' assessment of maximum employment. The second necessary condition was that we would have inflation recover back to 2%, and the third condition was that you were actually going to have forecasts for inflation depth to exceed 2% for a sustained period, so this was not going to be a one-time recovery to 2% and then back down, but there was actually going to be the expectation inflation was going to have some kind of makeup for past misses.
Amarnath: Those three were cast as necessary conditions. Some of them are clearer than others, like returning to 2%, that's something we can observe very clearly. We can look at the headline PCE deflator, see it's past 2%, it has recovered. On the other hand, where the FOMC sees maximum employment is more opaque. Where do I go to look up what the maximum employment assessment is for each FOMC member? Sometimes there are members who say it in their speeches, very publicly. Some are a little bit more coy. And some just have an understandably hard time grappling with the complexity of what maximum employment really encompasses. Is this something that changes over time? How do we evaluate over multiple indicators? We know the unemployment rate is flawed, can we do better? What kinds of metrics should we be looking at?
Amarnath: So that complexity ends up reducing itself in the dot plot and the Summary of Economic Projections as what is the Fed’s… what is my longer-run projection for the unemployment rate. Most members are somewhere around 4%, based on that. And so that's really all we have to go off of, but even if all the members have a different view about what maximum employment really is over time.
Beckworth: Do you think the dot plots are still useful?
Tweaking the Fed’s Dot Plots
Amarnath: It has a lot of information in it, but it's also very difficult to process that information. Because of the lack of transparency about whose dot is whose and what projection is associated with which dot, and your ability to extract reaction functions is very limited, and you don't know who are the voting members of the FOMC, who are not. That can really color how much information you can actually extract from it. We know that most FOMC members are expecting, if all goes well, that they will be expecting to raise interest rates in 2023. Is that because they expect inflation to be out of control in 2023, or they expect it to be out of control over a longer horizon and they need to tighten rates in order to avoid that outcome? Is it because they see risk management considerations dominating their rate path? Is it because of a Taylor rule? Is it because of different frameworks?
Amarnath: Different members have very different approaches. That heterogeneity is very hard to capture in the current iteration, as much as they're trying to be transparent and communicative, and I appreciate that as a Fed watcher. I also fully understand that this is complex, and it's not necessary that more information leads to clearer information all the time.
Different members have very different approaches. That heterogeneity is very hard to capture in the current iteration, as much as they're trying to be transparent and communicative, and I appreciate that as a Fed watcher. I also fully understand that this is complex, and it's not necessary that more information leads to clearer information all the time.
Beckworth: Well, we'll get to your big recommendations on the maximum employment front in just a minute, but on the dot plots, would you recommend any tweaks to it. So there's some FOMC members who listen to the show. They're listening now, Skanda, so would you maybe reveal who is each dot plot, what's behind it, or any minor tweaks before we get to the big ones?
Amarnath: So I think that the dot plot could benefit from a little bit more transparency. That may be hard to achieve, because there are other considerations about making sure it's a committee-wide view and not just Jay Powell's dot that matters. But I would say that Fed President Neel Kashkari's approach to communication is something I actually quite admire, of explaining where he was setting his projections and how he was looking at the interest rate outlook. In a transparent way, following the release of the projections and the release of an FOMC statement, that kind of transparency's really helpful, because otherwise we're left not knowing whose dot is whose and what's motivating that dot. I think this last FOMC meeting is actually quite illustrative of what the challenges of interpreting what is meant by two hikes in 2023.
Amarnath: Is what meant by two hikes the notion that there are inflation risks that are not part of my modal forecast, in which case I need to just hedge against the tail risk that inflation surges, or is it because I think maximum employment has already been achieved, inflation is slightly above 2%, and my baseline forecast warrants two rate hikes in 2023? There's some texture there that's hard to parse. Even if the Fed did a perfect job in that sort of release of the Summary of Economic Projections, if it's left somewhat opaque to the observer, markets move really fast, and markets are trying to process a lot of information, and it's not necessarily the case that every detail gets parsed perfectly.
Amarnath: So I think that just having a little bit more transparency about who's saying what makes a big different, because then you can actually account for it accordingly. To what extent does your actual expectation, mean expectation, compare with what you think is your baseline forecast versus the set of risks around it. Because most of the FOMC members do see risk to the upside to their forecast, that's also probably playing at least some role in their projections.
Beckworth: So we want every FOMC-voting member to have a Medium post after every meeting… That would be great. I'd love to go read all the blog posts from the different FOMC members right after the meetings and see where they're coming from. So it has been great, and I'll confess, I haven't been following the Neel Kashkari posts. So he's still doing them? I did early on.
Amarnath: I don't think he's... I've not seen him publish one more recently. I suspect there's a heavy dose of effort to kind of get those out every single time.
Beckworth: Yeah. Absolutely.
Amarnath: But I did think at least as a model, maybe there's a quick and dirty version of that, but I did think it's very informative of, "This is how I voted, this is what's guiding my decision making, this is what would change about my views if the economic outlook changed." Because look, part of this is the macroeconomy is going to change in ways we don't expect. What's really important is actually to understand how members read and react to the information, than to just say, "This is my forecast. This is my point-based forecast, and so just take that at face value." I think there's some additional transparency that would make the dots more effective, and in the absence of that, I can understand why Chair Powell appears to be frustrated by what they actually communicate.
Beckworth: Okay, one last question, I promise, then we'll get to your article. Some people have said that this dot plot from the last FOMC meeting in June undermined the credibility of this new framework, at least to some extent. Do you think that's the case, but it's not that big of a deal, we'll get past this, or is it more serious?
Amarnath: I think it is to be determined whether this has damaged credibility or not. You have seen, in market pricing, that there's been a noticeable upwards in front-end rates. Not a huge shift, but the market was of a belief that the Fed was going to keep rates low for a long time, and now there is a greater risk that the Fed's going to have to raise rates, and that's largely attributable to 2023 dots moving up. So we do have the projection of interest rate increases. Whether the inflation is consistent with that and how credible the Fed is about managing this process, that's a question of whether you interpreted average inflation targeting to be looking through transitory inflation pressures, or whether transitory inflation pressures actually counted towards the average inflation target.
Amarnath: I would say the transitory inflation pressures should not really be guiding monetary policy that works pretty bluntly, and monetary policy that is supposed to be targeted towards a forecast should be calibrated towards achieving forward outcomes. If we know that there's going to be strong price pressures related to reopenings, some local bottlenecks that are expected to abate over the course of the reopening process, and if we do have the kind of fiscal pressures that are, let's call it one time in nature. Let's say the transfer payments related to stimulus checks, unemployment insurance, or the kinds of things that are going to expire. These are not the things that monetary policy should be particularly sensitive to, they should be more looking at what's the outlook in 2022, 2023.
I would say the transitory inflation pressures should not really be guiding monetary policy that works pretty bluntly, and monetary policy that is supposed to be targeted towards a forecast should be calibrated towards achieving forward outcomes.
Amarnath: So in that case, the flexible aspect of average inflation targeting and the Fed's framework warrants not overreacting to these types of dynamics. But I think this is a part where there's probably room for multiple interpretations, whether that flexibility is to accommodate transitory pressures, or is that flexibility supposed to actually count in and say, "Well, now we've overshot to the upside because used car prices are being bid up by CarMax." This is a very different-
Amarnath: What is it that we're actually after? Because prices can move around for a lot of different reasons, and it's true, the price increases right now are very strong. Some of them are for reasons that are pretty predictable, like in the case of oil, and some cases are less predictable. That's not necessarily the same thing as what sometimes we talk about in terms of nominal GDP targeting and the level targeting aspect of that, which is really about keeping institutions, incomes and balance sheets somewhat more whole and more predictable as a result, which is not really reflected in pricing power in the same way.
Beckworth: Yeah, all of this has just reminded me that I like a nominal income target to avoid all of this confusion at some level. We'll come back to this point in a bit, but I think it's important that we understand these issues, because it's the world we live in. We don't live in a nominal income targeting world, but if we did, it seems like life would be a whole lot easier for Fed watchers and Fed officials themselves, given there's data considerations in implementing such a target. But let's move to your paper, again the title is, *Beyond the Phillips Curve: A Dynamic Approach to Communicating Assessments of 'Maximum Employment'.* As you mentioned, one of the big changes in their framework is that they're now looking at shortfalls from maximum employment, whereas before, they would look at any deviation, above and below. So big, big change, and you and Alex Williams have this paper out. I'm just going to read a sentence early on in your paper, and you say, "The Fed's communication with respect to its assessment of maximum employment is overdue for a clarification." So help us out, Skanda, why is it overdue for a clarification?
Clarifying the Fed’s Assessment of Maximum Employment
Amarnath: I think it is, in large part, informed by the same dynamics that informed the Fed's framework review and forward guidance, and that's the late 2010s revealed why trying to pin the unemployment rate at a particular level is not actually a good guide for how to manage demand-side policy. And so the Fed, from 2015 to 2019... Or 2015 to 2018, to be fair, the Fed had a tightening bias. They thought somewhere between 4.5 and 5% was where maximum employment was. They slowly revised down that number, even as the unemployment rate made faster declines over that period, and brought in new people through labor force participation. So there was prime-age labor force participation started to recover over that same period, and the unemployment rate was also falling, and it wasn't in any obvious way inflationary. I don't think we could say that the 2015, 2019, if you tell people that inflation was a real risk, I think you'd have had a hard time selling people on that proposition.
Amarnath: Certainly not more so than in previous periods. In fact, probably the strongest inflation readings of the 2010s were early in the 2010s, 2011 or so, because of commodity prices. So there is a recognition that trying to land the unemployment rate at a specific point isn't really the right way to go, and if you sort of stick to that number, what might seem as dovish policy at first turns into really hawkish policy, because you're trying to target an unemployment rate. But in fact, what we learned is, there's more that can be achieved over time. The time dimension, the temporal dimension is really important for how we think about the macroeconomic trade-offs. If there's more achievable as more progress is made, it's something that... I think Jay Powell's press conference, he had noted that the labor market had made continuous upside surprises in that period of time. The Fed, I think, does want to accommodate them, hence the asymmetric nature of their labor market assessments, for the Fed framework.
Amarnath: But as long as they say that 4% is really what maximum employment is, or that's what the public gleans from their Summary of Economic Projections, it's always going to seem as if, once we get to maximum employment, the rate hikes are right around the corner, and the Fed's going to try to manage the labor market aggressively from the upside. I think it would be clearer to say, "This is where maximum employment is today. This is where we think maximum employment will be tomorrow." Maybe that's more static in periods when the inflation risks are elevated, and maybe that's more dynamic in periods where the inflation risks do not appear as elevated, but the notion that we can't achieve higher rates of employment, especially adjusted for age and demographics, over time….
Amarnath: Well, the US is actually the outlier in this case. So if we look at prime age, 25 to 54, employment-to-population ratios in other countries, they have managed to make new highs. Particularly advanced economies, so I'm talking about Canada, Australia, New Zealand, Germany, Sweden. These are all places where you can find new summits, new peaks in labor utilization over time. That's not because of any particular structural reform, it has everything to do with the business cycle extended long enough for new heights to be achieved. Whereas, in the US, we haven't achieved those same heights, and I think there would be more room to do so if there was more supportive policy and a more expansive view of what maximum employment can be over time. So taking more aggressive revisions when we get closer to maximum employment on their estimates, that's to say that what the constraint is today is not the constraint for tomorrow.
So if we look at prime age, 25 to 54, employment-to-population ratios in other countries, they have managed to make new highs...Whereas, in the US, we haven't achieved those same heights, and I think there would be more room to do so if there was more supportive policy and a more expansive view of what maximum employment can be over time. So taking more aggressive revisions when we get closer to maximum employment on their estimates, that's to say that what the constraint is today is not the constraint for tomorrow.
Beckworth: Yeah, probably the most surprising thing about these projections, the Summary of Economic Projections or SEP, is that they look very similar to what they were before the new framework came into play. I mean, you would think a redoing of your framework where you have meaningfully changed the full employment definition would somehow be reflected in the SEP as well, but what you're pointing out is that it's the same old long-run unemployment metric they had before.
Amarnath: Yeah, it is, implicitly at least, a Phillips curve view that says the natural rate of unemployment is 4%. It's going to be 4%, and therefore, when it gets below 4%, we've got to lean against it. That may not actually reflect all FOMC members' views in a precise and nuanced way, but that is what the public can easily glean from the document. And so even if there are other indicators that different members think are better, if they have more expansive definitions such that they think the 4% number may not capture it… Nevertheless, when the dots are released, markets move. When the dots are released, public and media write their stories.
Amarnath: If part of the way the Fed's using the framework is to recalibrate how expectations are set about monetary policy, then I think a more dynamic approach to what maximum employment can be, and what it is today, versus what it could be tomorrow, or in 2022, 2023, 2024… These are all very path-dependent. They're all kind of sensitive to what was achieved today can determine how much more can be achieved tomorrow. It is not the case that there's a sticky level of employment or the sticky level of unemployment that, over time, defines the capacity constraints of the economy at large.
Beckworth: Yeah, so you're effectively making the case for endogenous aggregate supply or reverse hysteresis. Running the economy hot enough, at least to see if that's possible, right, if those frontiers get moved out some.
Amarnath: Yes, that's not to be a denialist about capacity constraints and supply constraints that may emerge in various forms at various times. Right now, I wouldn't say it's constraining the economy as a whole, but the semiconductor shortage is constraining the auto sector. Auto production is being held back mostly because they cannot procure the requisite number of microchips. In the case of the '70s and the 2000s... I think 2000 is actually a very recent example, where we did have elevated commodity price inflation, and that elevated commodity price inflation, with respect to energy and food, was the kind of thing that the Fed had to balance. We did have some inflation that was above target on headline, and in retrospect, above target on core.
Amarnath: That does color the trade-offs, and then capacity has to expand and catch-up, and that also has to be accommodated, in some ways, but it also has to be balanced against the inflationary pressures. What we have right now, though, is if you take the rigid Phillips Curve view, and you sort of say 4% is the condition for maximum employment... Which, Jay Powell says the maximum employment is below 4% in his... Or he's suggested that he hopes it's well below 4%. I think even that little bit of color is not the kind of thing that gets communicated through the Fed's formal communications associated with their dot plot.
Beckworth: Let me ask you, because you are a former hedge fund economist. So when that dot plot came out, was your instant reaction, as it affects trades for example, tied to the dot plot or what happens in a press conference?
Amarnath: I think, also, the timing is different, so their dots and the statement get revealed half an hour before the press conference. And so if you want to trade the press conference, you can trade the press conference
Beckworth: I see, I see. Okay.
Amarnath: Depending on your trading horizon. In terms of what tends to occupy everyone's attention, in terms of market actors, it's the dot plot.
Beckworth: It's the dot plot.
Amarnath: The dot plot is the thing. You can make all of the nuanced stories, "Well actually, the growth projections are revised up, and the inflation projections are revised this way. And actually, it's more hawkish." That's fine. I think this is probably what frustrates Jay Powell the most is like everyone is looking forward, did the median dot rise?
Amarnath: Okay, we expected it to be a 7-10 split, or a 9-7 the other way. I don't have to be in that game anymore. In some ways, it was fun, in some ways… I think it does start to get a little bit disconnected from what the Fed is actually trying to communicate. The two dots moving up was the story from June… Sorry, the two hikes. The median FOMC member now sees 2023 as the liftoff year. That's a story that's been priced into markets pretty coherently I would argue, and that's just something now we have to really say, "Well, why is that the case?" What are the economic outcomes that really align with that projection, and what are particularly dissonant and divergent from that projection for interest rate hikes?
The median FOMC member now sees 2023 as the liftoff year. That's a story that's been priced into markets pretty coherently I would argue, and that's just something now we have to really say, "Well, why is that the case?" What are the economic outcomes that really align with that projection, and what are particularly dissonant and divergent from that projection for interest rate hikes?
Beckworth: Yeah, so the SEP has become this kind of center of gravity for financial markets in terms of what to actually expect. It becomes something much more than was intended and obviously more than what Jay Powell wants it to be. I think it's your first point, and second point though, it really hasn't changed that much, right? Because markets are putting so much interest in making decisions based on the dot plots, we would've hoped to have seen something fundamentally different to reflect the new framework. Now you highlight in this paper, and you already have here, the full employment part. And you know my critique early on was the inflation part doesn't change much either. Until recently, it shows some catch-up, but even I have my doubts on the inflation front, I'll tell you, Skanda, because is that inflation above 2% simply base effects? Is it simply the pandemic? Now I know, depending on how you measure, it gets us back to 2% average, but I want to have a test run of this system in a normal period so I know for sure what I'm seeing is not base effects versus the actual policy at work.
Analyzing the Uptick in Inflation Numbers
Amarnath: Yeah, I think it is a communications nightmare, to be honest, because I think you have to bring up various different prices, and what's going on, and Jay Powell even talked about used cars and having to talk about specifically with hotel rates and airfares, and that is a lot of disaggregated understandings, and they're important. I actually think you do need to have some texture when you try to understand what's going on with inflation, in moment at least. But whether these actually sustain themselves… I think it'll be interesting next year in particular, when we've kind of come through this impulse, what's the inflation outlook really going to be if all the price increases were front-loaded but were going back to 2% or below 2% in 2022.
Amarnath: And if inflation sustains, I think that's probably an outlook where the Fed may feel more comfortable with their current projections, but I think it's also the case... And this is something John Williams said actually, in a speech recently, which was that, if we actually see all these price adjustments happen faster, we may not see the kind of inflation overshoot in subsequent years that... It depends on whether you think of 2021 as like, "Okay, we've overshot," or do we say, "Okay, actually, this was a one-time thing that is not really the product of the Fed's framework, it's just the product of we had a pandemic, we shut everything off… when they had to restock their inventories all at once, it led to some local bottlenecks." That's the other side of this. In which case, I don't think the framework itself deserves any great credit for leading to that inflation overshoot. You're back to square one of how do you reshape expectations.
Beckworth: Right, the accidental average inflation targeting Fed just happened to have a pandemic and a big fiscal package that kind of pushed it back to 2%. But yeah, that's the thing, going forward, is the credibility going to be there? Will markets understand this process, will the FOMC understand this process themselves? I think all of us are learning something as we try this new framework.
Beckworth: Let’s talk about your proposal. So there's a lot of issues. I think you've touched on some of them already. We need to know the difference between capacity constraints and labor utilization, is a point you bring up in the paper. Maybe just briefly tell us why is that? Why do we need to know that distinction?
The Significance of Capacity Constraints vs Labor Utilization
Amarnath: Because I think a lot of the constraints that are driving inflation are not always well cast in terms of labor market dynamics itself. Where you see the inflation may reflect the fact that the production capacity is deficient. I think one good example of this would've been in the mid-2000s. The bulk of the inflation in that period was not due to wage-price spiral. It was not due to labor being bid up or so scarce that it was driving the inflation itself, and wage growth was slowing through most of that decade, actually, if you look at the employment cost index. But in fact, it was because there was a shortage of capacity to produce the raw commodities that were part of the China boom, the EM boom, and the global commodity super cycle. It had some path through into core components, but when we describe the constraints here, then you have to really think about what is... Now, that may have implications for how to balance supply/demand trade-offs, that's true, but it doesn't say that there's a limit to what is an achievable rate of labor utilization over time.
Amarnath: It's one thing to say that there may be so much progress that's achievable today. If we try to go to a 1% unemployment rate economy in a day, and hire everyone, that's a lot of income that we're generating. That's a lot of income that can translate into spending for a particular thing, goods and services, and probably not be well-balanced with the stable prices part of the Fed's mandate. But it is through that over time more progress is achievable, and how we manage those processes is the kind of thing where what the summary statistic that I think some macroeconomists like to see themselves as macro, "I only look at the aggregated variable, don't tell me the sectoral story, don't tell me the disaggregated story." But it is really true that how you think about these capacity constraints, that's not a shortage of labor that we had inflation in the mid-2000s.
Amarnath: The reason we had price increases is because there was really strong demand globally for commodities, and there was a deficiency of inability to produce the necessary commodities there. With the advent of shale production methods, with advent of other investment in a lot of different areas, we had commodity over capacity for most of the 2010s. There was an overcapacity of copper, of oil, of a lot of agricultural commodities. So these particular dynamics, when you think about what drives inflation, that I think is a pretty critical point to make. Trying to say that because there's too much employment itself, or we can't achieve a rate of labor utilization beyond a certain point and that's a permanent ceiling, I think is misguided.
The reason we had price increases is because there was really strong demand globally for commodities, and there was a deficiency of inability to produce the necessary commodities there...So these particular dynamics, when you think about what drives inflation, that I think is a pretty critical point to make. Trying to say that because there's too much employment itself, or we can't achieve a rate of labor utilization beyond a certain point and that's a permanent ceiling, I think is misguided.
Beckworth: Let me dumb that down for myself here. It's a good point, but let me frame it maybe the way I would frame it, and I know I'm missing some of the granular points on this granular discussion. That is, it's the ability to see through supply-side shocks that affect inflation, and I know it's a little more complicated than that, but if I'm a macroeconomist looking at headline inflation, and I see inflation go up, 2008 is a good example, it's going up and it may be tied to capacity constraints, commodities, supply shortages, relatively speaking, then I might miss the underlying true trend of inflation. Is that the point you're making?
Amarnath: I think the true trend of inflation is probably better described in terms of income growth, but I think you're right that the supply shocks that are happening are not about the labor market itself being sort of overly hot. You can have inflation without having hot labor markets, and you can have hot labor markets without having inflation.
Amarnath: That is the part it takes a bit of parsing to get through, but it's true that these are two different phenomena. They have some relationship to each other, what relationship can be strong at certain points, and at some points it can be really weak. I think if you think that there is just a fixed level of employment that's achievable through business-cycle expansions, you are implicitly saying that trade-off is pretty stable. Once you get past a certain level of employment, the inflation risks emerge. This is actually the Paul Samuelson view of once you get to full employment, then all of the capacity constraints emerge, and you can't get past a certain point. Whereas the guts of this is we have to separate the two. There are supply shocks. They do influence inflation, but they may not actually be very informative for thinking about how to keep a stable path of inflation over time, a stable path of income growth over time, and sort of keeping labor markets reasonably tight over time.
I think the true trend of inflation is probably better described in terms of income growth, but I think you're right that the supply shocks that are happening are not about the labor market itself being sort of overly hot. You can have inflation without having hot labor markets, and you can have hot labor markets without having inflation. That is the part it takes a bit of parsing to get through, but it's true that these are two different phenomena.
Beckworth: I mean, the whole point of inflation targeting, and especially this average inflation target is to ground inflation expectations, right, over the long run. If anything, a credible average inflation target should give us more stability over the long run in terms of inflation, and we can forecast better, so a supply shock, a capacity constraint, we should definitely see through it if we are well-grounded in terms of the long-run path of inflation.
Amarnath: I think there's a good way to actually parse this even further, which is that 2008 and 2011, you saw commodity prices rise, but income growth was actually pretty weak. 2011, we were still in the aftermath of the Great Recession. It was still a really weak economy. Europe was brewing in turmoil. And 2008, we saw the housing crisis was also unfolding at the same time as the commodity price boom. And those certain circumstances, actually, the inflation outlook should've been guided more by what was going on with incomes and the business cycle, and not by commodity prices, supply-side issues. In the 1970s, nominal income growth was also high, which is something people forget about it. It's not just about prices itself. Yes, there were oil price shocks. Yes, there were ways it mattered, but it was also being reflected in the pace of income growth over a sustained period of time.
Amarnath: That is the part of it that, when we're thinking about how the Fed's supposed to be communicating what is a sustainable rate of income growth and a sustainable rate of labor market improvement, that's what has to really be separated out of it. These labor market outcomes can improve over time, the pace of that process may be slower or faster, depending on capacity constraints, and that has to be sort of taken into account accordingly, but it's not the same thing as saying, "Well, once you get past 4% unemployment, all hell breaks loose."
Beckworth: Yeah, so we've got to have a broader dashboard than 4% unemployment. In fact, let's use that to segue into your proposal. So you and Alex Williams come up with your version of the Summary of Economic Projections, so you modify it. Walk us through the changes that you would like to see incorporated into the SEP.
In the 1970s, nominal income growth was also high, which is something people forget about it. It's not just about prices itself. Yes, there were oil price shocks. Yes, there were ways it mattered, but it was also being reflected in the pace of income growth over a sustained period of time. That is the part of it that, when we're thinking about how the Fed's supposed to be communicating what is a sustainable rate of income growth and a sustainable rate of labor market improvement, that's what has to really be separated out of it.
Modifying the Summary of Economic Projections
Amarnath: In addition to the Fed's baseline outlook for GDP growth, for the unemployment rate, for core and headline inflation, I think it would be helpful to have a clear view, especially since policy is being anchored to this now, in terms of assessments and maximum employment. What are each member's assessments of maximum employment for 2021, 2022, 2023? Because those numbers could be different, they could be more similar, but it's actually not totally clear, especially as we move past the supply-side challenges of the current year, as more businesses reopen and go back to normal. We're going to probably see a very different picture in 2022 and 2023.
Amarnath: That could've been a one-way to really thread the needle on how the Fed saw maximum employment in the late-2010s, when we did get past 4.5% unemployment, and we weren't seeing the inflation from it. The Fed should've been knocking down their estimates of what's the sort of sustainable rate of unemployment over time. They should've been willing to revise that, and they could also accommodate those by saying there's more progress that's achievable as you move further in time, assuming no shocks, assuming no recessions. So taking a more dynamic approach, saying actually, "This is where maximum employment is in the current moment, but it doesn't have to be the ceiling in subsequent moments." That cuts both ways, obviously, but it should really cut in the asymmetric direction of saying these are shortfalls that we should be addressing today, but they don't necessarily guide us in terms of what maximum employment looks like a year, two years from now, in the same way.
Beckworth: Yeah, so I'm looking at your paper here, and you have, in this modified Summary of Economic Projections, you've got the existing long-run unemployment rate. You also have the employment-to-population for prime-age workers in there, and then you have the employment cost index.
Amarnath: Yes. And so I think there's something that the Fed talks about a lot, which is that you can't just summarize the labor market by one variable, and that's true. You can have high employment and not very strong wage growth, you could have low unemployment and sometimes somewhat stronger wage growth, so you want to capture that on multiple indicators. I think actually, from the perspective of really calling a labor market tight and hot, to quote Jay Powell, "You've got to see some heat. You have to see that there is the kind of wage growth that actually reflect a strong labor market." Are we seeing wage growth that is consistent roughly with productivity and broader nominal GDP growth? Are we seeing that there is that kind of bargaining power being exercised, and it's not just merely about getting to a certain level of employment.
Amarnath: We included the prime-age employment-to-population ratio, because at this point, the Fed really should be communicating labor utilization on a better indicator than the unemployment rate. There's some really good research about how the unemployment rate is so flawed that we don't actually do a good job of classifying who is unemployed and who is not participating in the labor force. The ability for surveys to accurately classify and consistently classify over time is not particularly good. This is actually a very blurry line. It's easier for me to say I'm employed or I'm not employed than for me to say whether I am unemployed or not participating. Respondents give very different answers at different times, and they're very confused about the definitions. As a result of that, I would think that the Fed, for that reason alone, should shift towards employment-to-population ratios as a way to communicate labor market health. You have to make the right adjustment for aging, so you can take 25 to 54, maybe something like 30 to 64, but there are ways to take a better age adjustment and still use an employment-to-population ratio to guide these judgements.
We included the prime-age employment-to-population ratio, because at this point, the Fed really should be communicating labor utilization on a better indicator than the unemployment rate. There's some really good research about how the unemployment rate is so flawed that we don't actually do a good job of classifying who is unemployed and who is not participating in the labor force. The ability for surveys to accurately classify and consistently classify over time is not particularly good.
Beckworth: Yeah, I think it'd be important for the Fed to sell this. I mean, it would be a new approach, right? Typically, we're used to the Fed doing the unemployment, as we've been talking about. I like what some have said in terms of the employment population for prime-age workers. That's a mouthful, so I've heard some call it the working-age employment rate. That's a nice, quick, succinct way of doing it.
Amarnath: I was going to say, there's definitely room for better branding. A snappier title than prime-age, 25 to 54 employment-to-population ratio. So any advice and suggestions on how to make it snappier, we’ll take.
Beckworth: But this is good, though, because you have three indicators of the labor market, and two of them are labor utilization, and the third is labor cost or labor wages, a measure of that. And I know that harkens back somewhat to both of our views on the desirability of looking at nominal income growth. But maybe talk about the ECI a little bit, why you included it.
Amarnath: Yeah, I think it's really important to have some guide for what income growth looks like in the economy. I think that's one part of it. I think it's good for both evaluation of the inflation outlook and informative about the outlook for maximum employment. It's like what the maximum is should be a function of what kind of bargaining power do we see exercised. It really also cuts through a lot of these structural unemployment rate discussions, where if we assume some people are structurally unemployed, we should expect to see that wages are going up for the people who are already employed in a more substantial way.
Amarnath: ECI is probably, even though it's released quarterly and not monthly, it probably does the best job of avoiding certain composition biases, so it's probably the best-in-class wage indictor we have, it's not without some flaws and head fakes, but it is probably the most robust between employment… So what you typically hear on jobs day, jobs Friday, is the average hourly earnings number. That is actually not a particularly great number of looking at wages. It can distort things in some pretty important ways, so if you lay off a bunch of your lower-wage workers, then technically wages go up for the average, and so it confuses average and marginal in really bad ways. Whereas ECI is a little bit more robustly measured and is trying to control for particular occupations and what kinds of wages are being paid over time. That would probably be the best guide.
Amarnath: The Fed, if you look at their green books, if you look at their Fed transcript materials, and if you look at the historic materials of how the staff forecasts. They do forecast what the employment cost index is supposed to look like over time. So in some ways the Fed staff's already got them started on producing forecasts for ECI should look like. Putting a maximum employment projection that would be associated with that does not strike me as particularly onerous, it strikes me as something that they're doing implicitly at least.
Beckworth: Yeah, so I encourage listeners to take a look at this proposed change to the SEP. We'll have a link to it up at the show page. So I want to switch gears here, so you've given us your concerns and your proposals to fix the full employment side of the mandate and the Fed's framework, but let's move to inflation. I know we've touched on it different ways, but let me ask a question about inflation just to start things off. We only have a little bit of time left here, so this will be the last part of our show. Does the Fed's new framework implicitly acknowledge that the Fed does not have a good working theory of inflation? I mean, isn't the whole approach to this framework, "We're going to wait until we see the white of the eyes of inflation, because our Phillips curve model did not do a good job predicting inflation."
Beckworth: Maybe they believe in the Phillips Curve, to some extent, as a theoretical construct, but they don't believe the measures they have or the inputs are good. Effectively, they're saying, "Look, we don't have a good working theory of inflation, so we're going to just wait until we see it." Am I being too harsh in my interpretation of that? Or is that fair?
The Fed’s Working Theory of Inflation (Or Lack Thereof)
Amarnath: I think that is pretty fair. I think, just to be clear, I think the Fed's view is not some sort of Phillips curve that is univariate. It is an expectations-augmented Phillips curve, and so they're looking at what their observed estimates of inflation expectations are telling them, and they're also trying to back out some assumption about what the unemployment rate would do to inflation over time. And then you do some risk management considerations on top of that. That would've been a pretty loose description of what the Fed's reaction function was prior to the framework review, or prior to 2019 at least.
Amarnath: Now, I think that they are waiting to see some signs of sustained inflation in a way that seems particularly replicable. That's the part that is hard to say, that the used car price increases of today are going to necessarily going to be predictive... It may not be used car price increases next year, but it may be a different set of prices the following year. It's certainly possible, but how you use that information is not actually clear. For as many instances of persistent inflation of the 1970s that are constantly evoked, there are a lot of examples of where we have had price volatility to the upside and the downside. That's also hard, because that doesn't necessarily sustain either.
Amarnath: I think the Fed is really searching for a good theory in some ways, and in a theory that they have to be accountable for, and that maybe those who are in other positions of influence, whether it be academic, or market practitioners, don't necessarily have to deliver on. It's a hard problem to resolve of how you keep yourself accountable to inflation on the upside in a way that's still managing business-cycle consideration and their dual mandate responsibly.
I think the Fed is really searching for a good theory in some ways, and in a theory that they have to be accountable for, and that maybe those who are in other positions of influence, whether it be academic, or market practitioners, don't necessarily have to deliver on. It's a hard problem to resolve of how you keep yourself accountable to inflation on the upside in a way that's still managing business-cycle consideration and their dual mandate responsibly.
Beckworth: That's why I want to see a nominal GDP level target one day, because it gets us away from worrying too much in the short run about inflation.
Amarnath: I think that's absolutely right, because you are seeing this now. It's a very turbulent time in the economy, for reasons that are very independent of the Fed, but implementing a new framework within it, you have to reference a whole set of prices that are idiosyncratic, and yet you're trying to make out what is the systematic component of all of this, and what matters, and that just seems so much cleaner if you are making projections and anchoring policy to what income growth looks like. If income growth is really strong in 2022, even without all of these sort of fiscal and reopening effects, if we see really strong income growth in 2022, I think that you can see the makings for more persistent dynamics. Right? But that's just something that's a hard judgment to make in 2021. As someone who did forecasts, forecasting is really hard, it's even harder if you're not really anchoring any of this stuff towards some view of what the income growth outlook is going to look like in nominal terms.
Beckworth: I agree, absolutely. So we need more work done on forecasts of nominal income growth in different sectors. All right, let me step back from all of this, because this conversation is kind of predicated on a view of inflation that is tied at some level to how much slack there is. If we have too much spending, and it's pervasive enough throughout the entire economy, eventually you're going to have inflation. So I guess my question to you is, what is your theory of inflation? We know the textbook Phillips curve theory of inflation. What would be your theory of what drives inflation?
Skanda’s Theory of Inflation
Amarnath: It is inherently both a monetary, macro phenomenon, but it is also a phenomenon that requires some assessment of capacity at a disaggregated level. So you could think of it disaggregated supply, aggregated demand, that might be my best way to put it, which is that you have to look at the particular capacity constraints that exist in the moment, and those are not easily summarized into a particular measure of natural rate of employment or capacity utilization at the aggregate level, but there were capacity constraints for which demand-side policy, sometimes, is not going to be the right remedy or can actually aggravate certain pricing power that's not actually resolving it.
Beckworth: Make it worse. Yeah.
Amarnath: Make things worse. If you look at the 1970s and the 1960s, we had a few things that were going on, and I think it's really important when you say there was a great inflation, there was also the great nominal income growth boom of that period. That was driven by not just strong household spending, but it was also strong capital spending, so real fixed investment was really strong during this period. So we will have some subsequent pieces on precisely the great inflation and making sense of it, because we had high inflation in the '50s and the '40s that subsided very quickly. Whereas, in the '70s, we did not have that kind of transitory phenomenon.
Amarnath: So when we distinguish what drives inflation, I would say it is, we have to be willing to make adjustments on the fly. We have to understand that some of this stuff is not visible by simple aggregate macro variables, and that means proceeding cautiously, somewhat humbly, and also having a view of, yes, supply constraints can emerge in different sectors in different times and affect the whole economy. The best we can do, at the aggregate level, is to really look at how are monetary and fiscal policy influencing income growth, aggregate spending growth, wage growth, labor market conditions. Those are things that are easily aggregatable. Saying that this is the point where you can't produce more things, this is the point where you can't, is not something that... I would actually say this is where you do want to have more information, more context, more knowledge. I think that's where the premium really is. That's more a short-form version. I realize this is a huge question.
The best we can do, at the aggregate level, is to really look at how are monetary and fiscal policy influencing income growth, aggregate spending growth, wage growth, labor market conditions. Those are things that are easily aggregatable. Saying that this is the point where you can't produce more things...I would actually say this is where you do want to have more information, more context, more knowledge. I think that's where the premium really is.
Beckworth: No, no, this is good, but you're staying in the here and now, if I'm the FOMC official, I need to get granular. Otherwise, I might make a mistake in my decision.
Amarnath: I think they all are getting granular. If you look at their speeches, they are trying to make sense of whether this is transitory or persistent. That is inherently a question about, do you think the price increases of today, like the information we have is really about composition. The Fed always does this, in a way, they're always trying to do headline versus core. The reason you separate out headline and core is because you think the non-core components are not very predictive of future dynamics, because ultimately, if the Fed reacts to every headline inflation dynamic and tries to curb inflation because of oil prices or food prices, they may not actually be responding to the thing that matter over the coming year or two years. Monetary policy should be forward-looking, it should not be purely reactive, because there are a lot of noisy bits of the information set that are not worth reacting to.
Amarnath: That's the same decision you're making right now. If you think capacity can quickly catch up in areas, then that's good, but if you think capacity is going to take a long time and it’s going to require a lot of investment and a lot of spending to happen, then the assessment does shift. I realize, I'm trying to summarize a lot of high-level ideas in a very short amount of time.
Beckworth: No, this is good stuff.
Ultimately, if the Fed reacts to every headline inflation dynamic and tries to curb inflation because of oil prices or food prices, they may not actually be responding to the thing that matter over the coming year or two years. Monetary policy should be forward-looking, it should not be purely reactive, because there are a lot of noisy bits of the information set that are not worth reacting to.
Amarnath: This is a very complex and rich problem. We will have something out, hopefully within the next month or two, specifically on these points, but I think it's a tricky problem I think the Fed is grappling with. The common criticism here is mostly about communication and making sure that's appropriately dynamic and flexible to match what the Fed really is doing, I think.
Beckworth: Yeah, we're definitely going beyond the scope of your paper here. I'll just leave with this comment, and I'll give you the last word, Skanda. I have not been someone who's been concerned about runaway inflation. I'm definitely team transitory inflation. Part of it is tied to my theory of inflation. I think more in terms of quantities, and particularly, government debt, in order to get sustained higher-trend inflation, you'd have to have year after year after year of government spending growth rates up, and I just don't see that happening. I don't think it's in the political cards. That's trend, now what you're talking about is more short term, but I, from the get-go, wasn't convinced that we're seeing a fundamental change in the structure of how much government is going to be spending going forward on a trend path. Now, that may sound a little bit like a fiscal theory of the price level, but I think, in my view at least, it's important to think about that, whether we get worked up or not about inflation.
Amarnath: We see this actually quite similarly, even though we come from slightly different perspectives, which is that we have to think about the generators of income in the economy. Is it government spending, is it private credit creation in financial intermediation? Are we seeing something that's going to structurally change between now and next year, after we get through the reopenings and the initial fiscal impulse of the American Rescue Plan, which again, you send people checks once. This is not the same thing as sending people checks every year. Now, there may be changes in the fiscal policy structure over the coming six months. It seems like there's a lot of big legislation being discussed that can change the path, so all these outcomes are policy-contingent. As a forecaster, that's important to recognize. Your forecasts are contingent on a certain baseline policy outlook, and if the policy outlook changes, so too should your forecasts.
There may be changes in the fiscal policy structure over the coming six months. It seems like there's a lot of big legislation being discussed that can change the path, so all these outcomes are policy-contingent. As a forecaster, that's important to recognize. Your forecasts are contingent on a certain baseline policy outlook, and if the policy outlook changes, so too should your forecasts.
Amarnath: That's something to be really mindful of as we go into 2022 and 2023. I think of these types of dynamics you're describing, like government spending. Maybe it's the case that now the private sector has a stronger structural appetite for spending, and capital spending in particular, that actually does kind of accelerate the economy in both nominal and probably real terms too. And then, you're kind of in a different world. But again, there's a lot of stars that have to align for that to be the case, and I think that's just where I would say “to be determined”, at least.
Beckworth: Well, with that great advice, our time is up. Our guest today has been Skanda Amarnath. Skanda, thank you so much for joining us again.
Amarnath: Thank you so much.
Photo by Olivier Douliery via Getty Images